基于构造Hankel矩阵的SVD陷波方法

Signal notch method based on Hankel matrix and SVD

提出一种新的通过加入引导信号构造Hankel矩阵经奇异值分解(SVD)滤除相应频率成分的陷波方法。根据待处理信号构造的Hankel矩阵,经SVD后其奇异值对应信号中不同频谱幅值的频率成分,提出加入某特定频率信号作为引导信号使得该频率成分成为信号中的主成分,形成易区分的奇异值对,在信号重构时除掉该奇异值对便可滤除相应频率成分。用本方法对脑磁信号进行50Hz工频陷波处理,达到了很好的陷波效果,且该方法不受传统滤波器陷波越深受影响带宽越宽的限制。

This paper proposed a novel signal notch method based on structure Hankel matrix and SVD by adding a guide signal. Used the original signal to structure Hankel matrix,the frequency components of the signal could correspond to certain singular values after the SVD transformation,and proposed a method to add a certain frequency signal to make the component become the principal one,then could eliminate the certain frequency component by removing the corresponding singular values. Lastly,achieved a good effect when used the method to process 50 Hz frequency trap of the MEG signal,and the method is not affected by the traditional filter notch deeper the more bandwidth restriction.